Remarks by Bill Gates
CEO Summit 2002
Redmond, Washington
May 22, 2002

BILL GATES:  Well, good morning.  I see you’ve all got your tablets; that’s great.  I hope not too many of you put your headphones on during my speech to listen to MSNBC or Web streaming.  Of course, you can call up the slides and make notes on those as I go along.

Well, as you know, we have this conference every year, and last year was particularly interesting because that was essentially the point at which people recognized that a lot of the Internet hype was not coming true overnight and were stepping back and looking at where does the industry go from here, will there be some type of quick recovery in capital spending, and what’s that pace going to look like.

Certainly the last year there’s been a lot more essentially sobering up around the very big investments that got made during those boom years.  Certainly the telecom space, the venture capital space and capital spending for IT equipment has stayed quite modest, as one would expect, with the economy in the condition that it’s in.

Now, what I want to share this morning is a view of how over the next five or six years there will be some very dramatic advances, many of those building on the investments that have taken place, the fact that the Internet is there, that connectivity to businesses and many consumers we can just simply assume that is going to allow us to do some very important things.

But what I want to suggest is it requires some breakthrough advances, that billions and billions are being invested in by the industry in order to achieve the dreams that were originally held about how these systems would be helpful.

A key point is that we're nowhere near where we should be in terms of how good these systems can be, nowhere near where we should be in terms of the ease of use, the kind of capabilities you get out of them and making it very inexpensive to have the very latest and the very best systems for all of your employees.

And yet even with the limitations that we’ve got, the impact on productivity has been pretty phenomenal.  If you just simply take the idea of Web browsing, e-mail and the productivity software that’s being used and that has had a very big impact.  People like Alan Greenspan and others are looking at these numbers and saying it really is a fundamental change and far better than they would have expected going into those boom years.

Well, are the underlying advances moving us forward at the pace that we need?  After all, it’s really the miracle of the microprocessor and many related technologies that got us the personal computer and got us thinking about digital approaches to all the different things that we do.  And we need more power, we need more capabilities.  Just thinking about all the photos that you want to deal with, and how you share those, requires a lot of bandwidth, a lot of storage that today’s systems don’t have.  If we want to get these processors to do better analysis and use some of that extra power for reliability, making us do less work, we’ll need more capabilities there.

Well, fortunately the basic science that drives these technologies is advancing.  Moore’s Law, that predicts the doubling in performance every 18 months or so, will certainly hold true over the next decade.  The improvement in the storage on these systems continues to be really unbelievable.  Actually, the storage capacity is going up even faster than the processor speed is improving, and so even things like storing video now becomes very feasible.  Of course, storing video raises the interesting questions about Digital Rights Management, which I’m sure will be discussed a bit at this conference, because that is one thing that’s not in place, an infrastructure that properly controls when and how intellectual property, including software, movies and music, how that should be used.

But the disks themselves are getting us capacities where even now videogames like the Xbox have a very large disk, and so you can do things like take the music that you’ve purchased, store that on the Xbox, and so when you’re playing a game you’re listening to the songs that you care about, not the same tune that happened to be programmed into that particular game.  So with this extra storage many new things become possible.

Network bandwidth is also pretty important.  This is an area where actually the supply of capacity, particularly in the long-haul area, exceeded demand enough that a lot of the bankruptcies of some of the telecommunications providers over this last year have really come from the brilliance of what the technology can give in terms of long-haul capacity.

Those increases in capacity still really matter, and so the fact that optic fiber bandwidth will continue to go up is very, very important.  That doesn’t solve the last-mile problem, which is the only area where you could say there’s a bit of disappointment that consumers won’t have all the hardware magic that they want, but that’s really the only one.  In all the other areas, like these wireless networks that we’re now finding in more and more corporations and more and more places that you visit, like hotels, convention centers where you might want to be connected up; those things are making computing more pervasive and making us more interested in having the right software run on these machines.

Now, how will this power reflect itself in the actual devices?  Well, of course, at this conference we’re engaged in a real experiment with the latest breakthrough in form factor, which is this tablet form factor.  As we’ve had these tablets out on various trials it’s been fascinating to see how people use them, because it’s not just the new capabilities of seeing an interesting article and writing a note and sending that off to a friend or going to a meeting and taking notes, or when you have your sales figures, when you see something that intrigues you being able to point to it and expand that data to really see what’s underneath, a number that surprises you as either large or small, it’s also things that aren’t pen related, for example, using photos or music or doing e-commerce.  The fact that you can hold it in your hand and it’s just there, you know, pick it up, use the pen, get some information about what’s available out there, instead of having to go over to your PC that’s at your desk, that makes a difference in terms of behavior.

You know, say you were going to tell somebody you wanted to share some photos with them.  Taking to your desktop machine is always inconvenient.  There aren’t two chairs there.  It takes a while to bring things up and you can’t really hold it and look and point at the different things that intrigue you.

With this tablet form factor, that becomes very natural; you know, a few clicks, bring up those photos, hand it to the other person so they can navigate through it or hold it between you just pointing those various things out.

And so a lot of the things that have held back the use of this technology, particularly at a consumer level but also for information workers, is the fact that the form factor is very limiting.

At Microsoft when we have meetings we have to decide is it a meeting where people are going to be having their PC screens open and typing on the keyboard, or is it one where we won’t have that.  And it’s a real trade-off in terms of the intimacy of the meeting, but having the devices there also can make things far more effective.  And so you really want to seek out the best of both worlds and that’s partly what this tablet reflects.

Now, the advanced technologies going into form factors other than the tablet, things like these new phones that have rich color screens, those have as much power in them as a PC had only two or three years ago.  And so the ability to get screen information there, even though the device is small enough to fit in your pocket, that’s going to be something that we just take for granted, that it will be common sense four or five years from now that everybody can call up their schedule, call up the list of people that they want to get in touch with and that they’re contacted on that device when there’s something important that they should know about.

We do need to get rid of a lot of the complexity.  You shouldn’t have to have many different phone numbers.  You should be able to have the software work on your behalf and decide whether interrupting you, that is ringing that phone, is an appropriate thing to do based on who it is that’s trying to contact you and how important it is, and if it’s not important enough to interrupt you, then saving the information and notifying you when that makes sense.

So a proliferation of devices, the pocket devices, the rich games; we’re actually starting to see the digital set-top boxes that have been a dream of Microsoft, an area we’ve invested in for many, many years, we’re finally seeing that intelligence actually getting into the cable networks so that you can do some interactive capabilities, have a rich guide, save your programs and watch them whenever you’re interested in that and even have interactivity where you can respond to an ad or something in the show, have things much more targeted to the individual, that infrastructure is now definitely getting in place.

I see two limiting factors to this.  One limiting factor is how natural the interface is, and the second is how effective the software is on these systems.

Let’s talk first about the natural interface.  Many of these things have been talked about for decades and decades, and here at this campus, where the vast majority of Microsoft’s R&D is done, we’re spending over $4 billion a year to advance these activities.  And in each case, major breakthroughs have been made where, using the power of the hardware that I talked about and using the latest software, we expect these new natural interfaces will be mainstream literally within the next several years.

First of all, let’s take speech.  Speech actually brings up the idea of what is the relationship between the phone on your desk and the PC on your desk.  The phone is very reliable:  You get in touch with people, you have that immediate conversation, a very, very important tool.  The screen, of course, is much better for getting you sales results, and things where somebody is not in touch with you right now, say, looking at a list of your voice mails; that should be very straightforward, having that integrated in with your electronic mail.

But in a sense it’s strange that the two devices are so separate.  We’re starting to see some pioneering work by us and many others where you bring the phone and the PC together.  So, for example, if you want to call up a potential customer and show them information about a product, not only can you talk to them, you have the information on the screen.  If you want to do remote training, where somebody’s learning something that can respond back with questions or take a quiz, the idea of having the audio connection through that phone network, but also having that screen active, makes a lot of sense.

And so over time a standard element of the PC will be essentially the microphone and the phone-like capability, and so the ability to initiate calls, have a log, take notes, but most importantly share your screen with that other person as you’re connected up makes a lot of sense.

Now when you have that microphone there, it becomes very natural to say, OK, not only do I want to talk to other people and share information, but I’d like to be able to give voice commands to the computer, I’d like voice annotations on my documents, but also voice navigation throughout the information that I have on this machine.  And so by bootstrapping it with voice, voice communication, voice annotation and the voice commands, we start to get a set of features that make the machine far more natural than it’s been before.

We’re starting to see this happen first in some of the Asian countries, particularly Japan and China, where the relative ease of using the keyboard for the alphabet that they have is much, much worse than the keyboard is for smaller alphabets like English or other Roman-based languages.  And so it’s the emergence there that really shows that this is coming into the mainstream.

Handwriting, of course, is a key element of the tablet experience.  Again, handwriting is like voice.  It doesn’t always have to be recognized.  If you want to send a message to someone and say “great work” actually in a sense sending it as handwriting is much more intimate.  It’s like taking out your stationery and writing a personal note instead of typing that note.  And certainly each person in their own personal way can take that great work message and kind of emphasize that you’re really trying to send a positive message, encourage the person that you’re sending the note to, or if it’s something you disagree with, I’m sure there’s some way with your ink to show that you’re particularly intent on disagreeing or asking the person to take a harder look at whatever is involved.

Ink is very natural in that whenever we have a document we circle things that we want to change, we put little notes in the margin. That’s natural behavior.  And it’s an everyday experience for me to be looking at a magazine article and say, one of my colleagues might find this of interest.  And I don’t just want to have them get the article, I want to have them get the article with a small note that it came from me and why I think that might be of interest.

But today it’s a very troublesome thing.  I can write up an e-mail and say go look at this particular magazine and read this article.  It’s really not worth the trouble, though, to create that e-mail and have the person have to go find the reference that I’m making.  If, on the other hand, it’s a simple matter of just scribbling that note on that screen, picking their name from my people list, which will be unified between the address book and the buddy list, just pick that and send it off, that makes a very big difference.

And so handwriting has got to come into the mix here, both unrecognized that you simply transmit to people as notes or annotations, and recognized handwriting.  The recognition algorithms have improved very dramatically, including the ability to deal with people who have different writing styles.  I’m left-handed, and so our tablet teams definitely have gotten a lot of feedback about making sure they can recognize left-handers.  It turns out it’s a tougher problem than recognizing right-handed writing, because left-handers tend to cross their “Ts” a lot later.  They come back and do it. And the computer is always anxious to get going, so it tries to recognize the word before you’ve gone back and crossed the “T”, which, of course, often leads to mis-recognition.  So there’s a lot of cleverness built in there and over the next few months, before the October launch, we’ll be refining this and refining this based on feedback, including all of you.

Another big issue in making things natural, making it obvious to use the digital approach is the display.  For readability you’ve got to have two things.  You’ve got to have something you hold in your hands and you’ve got to have something where it’s as good as paper in terms of what the text looks like.

And so particularly the LCD technologies that some others have improved a lot, when you’re moving around the size of the display you want is limited by what you can hold in your hands and so the term tablet is very appropriate there.  When you’re back at your desk our view is that you want to deal with a wide range of information and you’d want even more of a display.

Now, that display should be something that you can point to information and deal with in a very direct way, kind of like the tablet, but it will look a bit different.

And I’d like to ask Gary Starkweather from Microsoft Research to come up and talk to us a little bit about what’s going on with these new displays and what is a desktop likely to look like over the next five years as we give people a chance for the best productivity.

GARY STARKWEATHER:  Thanks, Bill.  Thanks very much.

BILL GATES:  You’re welcome, Gary.

GARY STARKWEATHER:  It’s good to see you this morning.  It’s nice to see all of you as well.

What I’d like to talk to you about just briefly is the use of large displays, so we’ve built a prototype here that will allow us to investigate how users get to use large displays.

Two things with this:  First of all, it’s curved.  This means that the user can become engaged, become immersed in the work that they do.  And if you’ll notice, you see it looks like three monitors here, and it really is three projectors but nevertheless it can be put together as a seamless interface.

So, for example, we have called this particular technology D-SHARP. It stands for a Display System Using High Aspect Ratio Projection.  But what you want to see is the kind of capabilities you can have for putting up an entire picture here.  So what we will do then is to give you an example of something that you might find particularly useful. 

So here, for example, is an Excel spreadsheet.  There is the D-SHARP logo on top of that, so you can intersperse applications if you want, but additionally you can also have an application here for Word.  So users might have multiple applications up in which they don’t have to housekeep to put things away but nevertheless they can engage those as they want.

Now, one of the interesting things is you might enjoy seeing the whole spreadsheet so you can pull it open and of course get spreadsheets that you can get everything onto, which is really the intent.  It reduces the amount of mouse motion, the amount of housekeeping requirements to engage all the data.

So you can have that.  That gives you the ability to use the whole screen or just to use a portion of it, depending on what you want to do.

If we then put Word over here typically one would edit it page by page, but with a big screen like this you can clearly put up a five page spread and therefore when it comes to editing something I can come over, for example, to this graphic and I can move it over to some location like this and watch reflow occur before my eyes, and therefore that reduces greatly the amount of switching one has to do to find out where the information went, finding page 2, maybe printing it to see the results.  And if you say, “Well I really don’t like that,” you can just go put it back over here and watch reflow occur again.

So this gives you the ability to have a much larger workspace, it’s more natural for you to do this, as well as you have this engaged experience to handle this whole process.

So what we’ll do is we’ll put this away and then just from the standpoint of pictures you ought to understand one other thing and that is you can do serious photographs on here if you wanted to do your vacation photos.  You can have sorts of things that really become much more engaging in the capabilities you have and indeed we think if you do this correctly you can probably almost get the whole world on here.

Thanks very much.

BILL GATES:  So I said two things that were key. First is the natural interface, and we talked about that, and the second is breakthrough improvements in the software. And when we think about the software doing a better job, I think it’s worth breaking that down into three areas that are incredibly valuable in terms of how software helps businesses get their job done:  The information worker area, the IT department and how they have to build and maintain and secure these systems, and then the overall business processes and how those work inside a company.

Now, in each of these areas, what I want to suggest is that what we have today is really falling short and yet there is a particular approach that is going to make a huge difference, that is, move us up to a whole new level of what people will be able to get out of these systems.

I’ll start with the IT department.  They’ve been handed in a sense a tougher and tougher job as time goes on.  Originally they had mainframe systems and everything was contained inside the data center.  The total number of systems they had to think about even in the largest company would only be in the dozens, and so the idea that you could visit each one of those systems, have experts for every one of those things, it was all very plausible and you didn’t have outside connections where people were able to come into that mainframe and try and essentially attack the mainframe or do anything with it; it was truly isolated both physically and logically and it ran only the applications that the IT department put onto it.

Now, that changed very radically first with the arrival of the PC.  The PC meant that these people first of all got pulled into information worker productivity, things like e-mail, is the system working, can the documents be exchanged, and the scale of this was quite dramatic, thousands and thousands of machines, and they didn’t have the same control that they had of those mainframes.  After all, these were the tools that were allowing well-paid, important people in the company to get their jobs done and so in terms of what applications those people had used, how they would use those things, there was always they had to accommodate what was interesting there.

And the software that ran on these machines was typically designed to run on that individual machine.  The software got better and better and better, but there was no way to have a piece of software that could easily control and manage every one of those machines.  And so a lot of the tasks that the IT department would do would be proportional to that number of machines or how those machines work together.

At the same time the server challenge for IT also got a lot larger because the lower cost servers that were storing departmental information, and now all these servers that are the Web sites for both business interactions and letting customers come in, those proliferated as well and so you’ve got thousands of PCs and hundreds and hundreds of servers.  And so keeping these things up to date is a challenge.  Being able to have visibility of what’s going on with all of these different systems is a challenge.  If somebody is confused about something that’s taking place, that’s a challenge.

And yet in the face of that, at the same time as that, the expectation of the reliability of the systems also went up dramatically.  When it was the data center you could print the checks an hour late without a big impact.  Here, if e-mail is not up and running every minute, if the Web sites not up and running every minute, that has immediate impact in terms of productivity and customer perception.

And so from the IT department’s point of view, all these different systems and pieces of software have required them to come in and have the expertise essentially to try and pull those things together, and there’s an incredible number of boundaries that they deal with in this.

So that’s the first domain where although things are working, it’s far short of what should be possible.

Now let’s think about the information workers, and are they getting exactly the information they need to make them as effective as they’d like to be.  Well, of course, jobs of information workers range quite a bit from the very structured job of somebody who’s in telemarketing, or who’s involved in purchasing, all the way up to the managers whose jobs are much more ad hoc, and the CEO being the extreme there, where your focus really depends on the latest events that are taking place and you’ve got to have a lot of information just to decide how to prioritize what you want to do.

If you think of the software in your companies today and think about, you know, say you want to pull together a meeting on short notice, is it as easy to do that as you can imagine that it should be, where you have visibility of people’s schedules and the importance of the different things and the ease of moving those things around?  You know, I’d say that even in the most aggressive companies, the software is falling far short of making that simple.

Say you want to sit down with a group and really understand where their resources are, say you’re asking them to do a job maybe with less headcount, or trying to make sure that the additional headcount goes into the right places, what kind of visibility do you have of the organization, the results coming out of those different groups and the relative productivity of groups within your company and other outside companies?  You know, I’d say very limited visibility.

Looking into sales data and spotting trends, having all the IQ of the organization able to look at those numbers and probe down in and share insights about those things, you know, another very tough problem.

Getting alerted to things when it’s important, instead of having to go out and poll for information or getting things that aren’t as critical, again, the software infrastructure isn’t there for that.

Now, if we look at a more structured job we might think, okay, well this is a simpler problem.  Purchasing -- and I’ve picked the purchasing department as an example because that’s one where Microsoft has gone out and done a lot of essentially time and motion type studies, where dozens of these purchasing agents were nice enough to let us literally sit there and watch what their day looks like and look at how they’re using software, how they use phone, fax, e-mail, spreadsheet and is the job of that purchasing agent in any size company, and we did companies of many different sizes -- has it really been automated to the degree that it should be?

And it’s pretty stunning; the first few hours you sit there and watch, you realize that they’re having to deal with so many different types of information, so many different sources of information.  Some of the people they work with use EDI, but most of them do not.  Many of them have e-mail systems, but some do not.  They’ll often fax something out and just have somebody fax it back to them with scribbled notes on it.  And they’re dealing with many things all at a time, they’re trying to split the tasks between different people in the department; tracking what’s going on is very tough. And a lot of that job is still very much involved with writing down on paper notes about the things taking place and trying to keep those notes organized. It’s not done digitally.

We actually went in and built a diagram that shows all the different steps that these purchasing agents have to go through, and these steps, some of them can be automated.  Some of them are purely mechanical things like order confirmation or things like that, and some of them are not purely mechanical.  Some of them involve lots of human judgment where they know that if one supplier says that something can’t be done who might be able to come up with that, meeting that need, how they might be able to talk with the customer on the other side about how that works, and so they’ve got to be personally involved.

And many of these attempts to digitize things attempt to go too far in the sense of making everything purely happen on the computer, and not involve the human skills when that’s appropriate. 

And so every time you have a boundary where you put things into a software system, and yet you want to have visibility with the person, that’s very difficult for these people today.  And so as we just looked at how they went about their tasks and the sharing and the different things involved, it was clear that there was probably a factor of improvement that could be had if everything was done with software information that was brought together in a simple way.

The final domain I mentioned is the idea of these business processes, these applications inside the company that deal with things.  Just a mundane example would be a new employee comes in and you have a variety of things that should take place.  The person should get various permissions to see different files.  They should be set up in the mail.  There are probably a dozen applications that relate to employees, whether it’s the payroll or the purchasing, and you want all that information to be set up very automatically, and as that information changes, as the person moves into different roles, you want it to all be kept very up to date.

That is today done in quite a manual fashion.  The different systems, which were selected for their excellence, are all self-contained, and so there’s no way in today’s approach that that information flows easily between the different systems.

Another good example of this boundary problem in the software applications is thinking about customer feedback.  You have sales management systems, shipping systems, you’ve got trip reports, and so again the information is scattered around and very difficult to pull together.

I was looking at last year’s CEO conference and saying, well, there were some pearls of wisdom that came out of that.  One of the people we had was Jeff Skilling, and he was saying, “Well, you know, things are moving so fast, maybe you don’t need to pull all this information together.  You don’t need strategic planning.  Just go full speed ahead.”  Well, maybe you do need a strategic planning department and pulling all the information together.  Of course, it depends on what your goals are, but it’s an interesting insight.

So why isn’t software stepping up to these problems?  After all, software is this magical stuff that can automatically make things happen. It can take things that are manual and take over those things.  And I’m sure many CEOs have sat with their IT people and brought this kind of commonsense view that this should be straightforward to the IT department, to say, why is the budget going up so much relative to the kind of automation, that should be possible.

So why aren’t all those things I just went through, why aren’t they solved?  Why can’t you get the benefits of picking the best approach in individual areas and yet having the information across different areas come together in a rich way?

Why aren’t these systems that have all that speed that I talked about and all that capability, why can’t they essentially be self monitoring?  Why can’t they see if something isn’t working, and have enough redundancy that when something isn’t working they reset the system that isn’t and have other systems take over?  You ought to be able to use that extra capacity and apply it towards incredible reliability.

And why do all these systems feel like in a sense there’s more and more complexity, whether that’s at the end-user level, all the different commands you have to learn across different things or the IT department and how they describe the challenges that they’re facing?

Well, there is one systematic answer to this and it has to do with the way these things are architected.  Everything I said relates to a very common problem. It’s about boundaries.  When you have multiple applications, those applications don’t see each other and things don’t flow. When you have multiple systems, say, you have a PC at home and a PC at work, if you update your address book at one, it doesn’t happen in the other.  There’s no connection there and you manually are involved in pulling that together.

Between software and people, the information worker is taking those phone calls and knows what changes need to be made, but they often are very frustrated about how they get that information into these different computer systems.

And certainly between organizations it’s very tough because of security walls and authentication things, the idea that your partners and you could have common documents that you're sitting on and editing together and sharing those things, that’s very tough to do.

The best that can be done today is simply using electronic mail where you’re just mailing out things, and you get various people proposing edits on those things, and you’re trying to pull it back together.  There’s no real sharing there; there’s just e-mail going back and forth.

And so a lot of these problems come from the fact that the software was designed for the user to sit in front of the software and do something instead of the software itself being able to be controlled by other software.

Now, the technology that is changing this, and we’re actually through the early years of this work, we're through the years of proving the feasibility and getting the enthusiasm there, and we’re just at the point where we’re starting to have all the leading edge companies benefiting from this, is this so-called XML approach.  The standard actually got discussed back in ’97.  It was issued officially in ’98.  We decided for all our software to re-architect it around this approach several years ago and announced that as what we call the .NET Strategy in June 2000.  It’s a huge impact, because every piece of software has to take this new approach, the Windows system, the database, the Office software. 

Also, one thing that was a little scary about going down this road is if you go down this road, and only one company goes down this road, then it’s really a dead end because we don’t do the applications, and there are many other software stacks from IBM and other companies that people need to work together in those scenarios that I talked about, so it only makes sense if it becomes something that’s industry wide.

This year that’s really happened.  This year is the first year I can say for sure that the leading companies have decided this approach they’re going to build around it, and build around it in a way that actually allows for the interoperability; that is, if you want to do rich transactions between your company and another company using these approaches, if they’ve got, say, a pure IBM stack and you have a pure Microsoft stack, those things can interact or work together.  Or even if you mix and match inside the company, where you have different applications, those things can come together.

What it means is that the competition is not on the actual standards themselves, but rather building the tools and the implementations, the systems, who can build those that are the cheapest, the most reliable, the highest performance, and just like we saw with the PC itself having a framework where you have standards, in the case of the PC, the standard of how the software and the application was written, that can create phenomenal improvements in the price performance very rapidly because it frees up in that case the people who built the PCs to innovate and compete, and the people who build software know that their investments will run on all those different machines.

So let me make it clear why this is so radical and why it addresses what seems like a very broad set of tough problems.

Before you have this approach, the information inside a piece of software was inside it, and so you can see this is a purchasing application.  It’s got the prices, the catalogues, the orders all inside.  So the application was designed so that a human would come and see various screens and give various commands, and that was all that could happen.  If another piece of software wanted to get at that information, there was no standard way of doing that, so only the user sitting there giving the commands could have that interaction.

Now, when you take and adopt this approach, you actually take the information and you not only expose it, but you have a description of it so it’s easy to find and it says exactly what form it’s in so that another piece of software can come in and take that information and look at it.

One way that this has been described, that I think a lot of people understand fairly well, was to say that you’re essentially putting a ticker, an information ticker, on the outside of that software and so other software can gather the information.

So what does this mean?  Well, it means that when you have those thousands of PCs, instead of worrying about each one individually, you have a piece of software that can go out to each of them and say, you know, what’s happened on that machine, is it responsive, what’s installed, what version of the software there is, and can automatically for all those machines bring them up to date -- and that’s without anybody having to go in and visit those individual machines.

For something like doing business across the Web, what this means is if somebody wants to buy your product you have on your Web site, this kind of information exposed so they can see what products you offer, they can see what kinds of terms are there.  If you want to have customer references up there, they can see that information and their software can interact with it. They don’t have to manually go in and, say, look at every Web site of everyone that they might want to do business with.

And so in a sense it’s a very simple idea, but it’s a very profound idea, and it’s actually taken the last four or five years of work of companies like Microsoft and IBM and others to pull together this basic approach of how you expose that information.

It was a few months ago that a group called Web Services Interop, WS-I, was announced, which is the body that will make sure that these things really work together.  There’s a very specific commitment by Microsoft and IBM to demonstrate exactly how all this work we’re doing works together, and so that people can simply mix and match without giving up the benefits of these things.

Now, this approach is very interesting because it’s in contrast to what people were talking about three or four years ago.  It was actually at a CEO conference a couple years ago that people were saying, “Hey, these marketplaces, maybe those are the big things,” and asking whether they should join up and should they give exclusives to these B2B marketplace companies.  I spoke up and said, “Well, because this is the Internet, you shouldn’t have to only go through a particular middleman who can mark up the transaction and be in a very unusual position with respect to that market.” You know, for a while there were companies who were going to be the marketplaces for a market, where the value of that marketplace company was greater than the value of the people who were going to use the marketplace.  And so clearly that valuation suggested that there was going to be a lot of cost, a lot of friction in having to go through essentially a middleman there.

Now, why were people even proposing that approach?  It was because there wasn’t this XML standard that would let any two companies do business together without going through somebody who massaged the data and put it into a common form. 

And so the idea of the marketplace, there was a reason people were proposing that.  It was to affect the standards.  But now those standards, both the broad standards that are used for all types of companies and the industry specific standards for XML, for example, what does a patient record look like or what does a portfolio look like, those have been pulled together.  And so, although in some cases these marketplaces will have unique value added, the fact is any business now who’s empowered with this XML software can do business with whomever they want. 

And that really fits with what we’ve heard as we’ve gone out and talked to these businesses.  Every business thinks of themselves as a hub. They deal with larger companies and smaller companies, and they know there’s not uniformity in terms of technology adoption in the different people they work with, and so they have to in a sense take the older approaches and have software that works with those, as well as have software that works with the newer approaches.

But the beauty of what’s going on now is that that company, by buying software -- very, very inexpensive software, costs less than the hardware that it runs on -- they’re empowered to do this without anybody being in their transaction flow.  It’s their transactions, it’s their information.

And so although there will be B2B marketplaces, they will exist purely not because they got some big exclusive; they’ll exist based on the value-added that they bring, and the buyers and sellers will always have the option because of these standards of doing business directly, doing business one on one.

And so in a sense you could say these XML approaches were invented to solve an important problem, that is, that when the Internet connections came along and there was browsing people wanted to do more, they wanted to do buying and selling.  And this XML approach is exactly the design that’s necessary for that to happen.

But the really stunning thing, the amazing thing, is how that basic approach isn’t just useful for this business going across organizational boundaries.  It’s very profound in terms of how the applications within a company share information, how the information workers share information back and forth between those applications, and even how we manage these systems, how we have visibility and control and can start to say, OK, now the complexity curve has reached about the level that is the maximum, now we can use software controlling software to manage these systems so that the reliability and the manageability is dramatically different than it’s ever been before.

I mentioned that this is actually very concrete stuff.  It’s not something that’s way off in the future.  For us a big milestone was our developer tool that we put out early this year, which was the first major rewrite of a development tool to make it easy to write XML type applications. It’s called Visual Studio .NET.

We actually had people who were enthused enough about it even in its early testing period that they wanted to go ahead and deploy applications based on it.  An example of that is Traveler’s Group, and in their insurance claim system now there’s a really radical change that’s taken place, where they eliminate a lot of paperwork and they let these suppliers, for example, people who do glass repair, go in and get the confirmation so they can go ahead and do the work for the customer without any delay at all.  The car comes in, they connect up to the Traveler’s system, get that authorization and so the work starts out.  The total time to process the claim, because this is reduced by 30 percent, actually the overhead and cost goes down as well.

This is also a good example of the fact that they didn’t have to take all their software and rebuild it in this new approach.  They actually took the mainframe software that they didn’t want to change and simply put an XML layer on top of it -- sometimes that’s called the wrapper -- and then so the newer application could interact with that.  And so it wasn’t a rip-and-replace thing; it was just the new functionality, and then that wrapper let it interact with the claims database application that they wanted to leave the same.  Now, over time that’s something they may want to rebuild and pull together and get even more flexibility out of it, but they can take it a step at a time.

And so the real message here is to share my excitement about what’s happening in the software world; that is, software using this new approach that will let us, as we move forward, not be on the classic trade-off curve of, OK, the more function you want, the more complexity you get; the more function you want, the more manual effort there is to go out and do these things.  And it’s really created a tension where you’ll often have your information workers who want to be empowered saying, “Give me this new thing, this thing is great, I love this,” and then the people who have to make it all uniform and make sure it runs all the time pushing back and saying, “Boy, because we have to go in manually and do these things, it’s going to be quite expensive, let’s not necessarily jump into this or not jump into this at full speed.”

There are a lot of things that have driven the industry to this point.  Some of the e-commerce opportunities have gotten us here, some of the amount of glue code that’s had to be written between these systems.  Many people in computer science who have been working on ideas over the last decades can point to this and say, “Well, we’ve been predicting that this would happen.”  But in order for it to happen, it required many billions of dollars of research.  Even over the next couple of years we will have gotten so that we’ve spent over $10 billion in total ourselves on taking our software and applying this new approach.

And as I said, it’s not just Microsoft and it can’t be.  If you take industry wide as this is happening, and there you’d want to take a longer timeframe, because some of the people are just coming along and doing these things, industry wide it would be many, many tens of billions.  But it is something that’s quite dramatic. 

What it suggests to me is that the investment level that people have been making in their information workers, if they simply keep that at about the same level so that the wireless networking can go in, the network can be maintained, that many of the aspects of managing the software can become less costly and so there’s room for neat new things.  Some of those you’ll hear about a little bit more tomorrow with Jeff Raikes talking about digital meetings and real time communications, and those won’t require an up tick beyond what’s there because of the advances there.

We will see people using the PC technology, which now can provide the very high performance and reliability, to build these new XML applications and get the great price performance with the high demands they have there.

As I said, within every industry it’s interesting to look at the pioneers who are already pushing this forward.  There are a few industries, particularly financial and some of the healthcare, who have been pushed to it most rapidly, healthcare because of the need to do the digital patient records, now financial because they’re always really pushing the limit in terms of great information management.  And so there are some fun successes coming down this line. 

The job that I’ve got, my focus at Microsoft is really spending virtually all my time making sure this gets out there and shaping it into reality.

Thank you.